In optical communications and optical switching it is well known that signals can be transposed from a first optical signal of a first channel or wavelength to a second optical signal of a second channel or wavelength.
A detector remodulator may be used to convert the first optical signal to the second optical signal and involves the detection of the first signal in which the first (modulated) signal is converted into an electrical signal, followed by the modulation of light of a second (unmodulated) wavelength/channel by the (modulated) electrical signal. Whilst in the electrical domain, the signal may advantageously be processed, for example by one or more of amplification, reshaping, re-timing, and filtering in order to provide a clean signal to be applied to the second wavelength/channel. However, currently in the art, to amplify and filter the electrical signal at high data rates with low noise, the circuitry must be contained in a separate electronic chip, which requires packaging and mounting thereby increasing size and cost and reducing power efficiency.
In U.S. Pat. No. 6,680,791 an integrated chip is provided with a light detector and modulator positioned close together so that the electrical connection between the detector part and the modulator part is short and of low resistivity. However a maximum of only 10 Gb/s data speed is predicted for this structure due to diode capacitance and thin-film resistance limitations [O. Fidaner et al., Optics Express, vol. 14, pp. 361-368, (2006)].
U.S. Pat. No. 6,349,106 describes a tunable laser, driven by a circuit with a signal derived from a first optical wavelength. However because it comprises a III-V-material photonic integrated circuit and involves the use of epitaxial heterostructures and a vertical p-i-n diode structure, is inflexible in its design and therefore inadequate for new applications involving increasing switching speeds, reduced latency, reduced power consumption and the demand for lower cost and high-yield manufacturability. In particular, because the semiconductor devices including the modulator built upon the semiconductor chip are driven by circuits completed between contacts on the top surface and a contact covering all or a large proportion of the base or underside of the chip, the capacitance of the device cannot be readily controlled by design features built into the structures such as doped regions and metal contacts.
Application of arrayed waveguide gratings (AWGs) in optical switching (including optical circuit switching and optical packet switching) has been slow despite the advantages that AWGs may provide. This is in part due to disadvantageous features of AWGs such as the uneven response across a range of wavelengths, cascading effects of limited bandwidth after multiple passes as well as cross-talk between different ports.
Use of AWGs in optical switching is known; Ye et al (IEEE/ACM Transactions on Networking, VOL PP, Issue 99, Page 1, February 2014) and by Bregni et al (IEEE Journal on Selected Areas in Communications, VOL 21, No 7, September 2003). Ye et al describes the use of AWGs in Clos-type optical switches and other architectures and Ngo et al (Proceedings 23rd Conference of IEEE Communications Soc, 2004) has illustrated AWG switch architectures that are rearrangeably non-blocking and strictly non-blocking.
One of the difficulties in realising optical switches is speed and another is latency. Poor latency is especially undesirable in uses such as high performance computing and datacenter switching where it is desirable to make calls or rapid data exchanges on a system as close to real time as possible.